[ 
https://issues.apache.org/jira/browse/HAMA-367?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13041093#comment-13041093
 ] 

Edward J. Yoon commented on HAMA-367:
-------------------------------------

Please see the below code block. 

{code}
    while (it.hasNext()) {
      Entry<InetSocketAddress, ConcurrentLinkedQueue<BSPMessage>> entry = it
          .next();

      BSPPeerInterface peer = peers.get(entry.getKey());
      if (peer == null) {
        peer = getBSPPeerConnection(entry.getKey());
      }
      Iterable<BSPMessage> messages = entry.getValue();
      BSPMessageBundle bundle = new BSPMessageBundle();
      for (BSPMessage message : messages) {
        bundle.addMessage(message);
      }
      peer.put(bundle);
    }
{code}

Here's my suggestion for you.

1. write messages to file somewhere in local disk. 
2. Create compress/decompress methods
3. Send

> Runtime Compression of BSP Messages to Improve the Performance
> --------------------------------------------------------------
>
>                 Key: HAMA-367
>                 URL: https://issues.apache.org/jira/browse/HAMA-367
>             Project: Hama
>          Issue Type: New Feature
>          Components: bsp, documentation 
>    Affects Versions: 0.3.0
>            Reporter: Edward J. Yoon
>            Assignee: Edward J. Yoon
>              Labels: gsoc, gsoc2011, mentor
>             Fix For: 0.3.0
>
>         Attachments: test_files.tar.gz
>
>   Original Estimate: 2016h
>  Remaining Estimate: 2016h
>
> As you know, the exchanging data between processes, is a core part of whole 
> performance in Bulk Synchronous Parallel.
> In this research, we investigate BSP message data compression in the context 
> of large-scale distributed message-passing systems to reduce the 
> communication time of individual messages and to improve the bandwidth of the 
> overall system. 

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

Reply via email to